Paper: arXiv 2505.11163

Authors: Anubha Goel, Puneet Pasricha, Martin Magris, Juho Kanniainen

Abstract

Time series foundation models (FMs) have emerged as a popular paradigm for zero-shot multi-domain forecasting. These models are trained on numerous diverse datasets and claim to be effective forecasters across multiple different time series domains, including financial data. In this study, we evaluate the effectiveness of FMs, specifically the TimesFM model, for volatility forecasting, a core task in financial risk management. We first evaluate TimesFM in its pretrained (zero-shot) form, followed by our custom fine-tuning procedure based on incremental learning, and compare the resulting models against standard econometric benchmarks. While the pretrained model provides a reasonable baseline, our findings show that incremental fine-tuning, which allows the model to adapt to new financial return data over time, is essential for learning volatility patterns effectively. Fine-tuned variants not only improve forecast accuracy but also statistically outperform traditional models, as demonstrated through Diebold-Mariano and Giacomini-White tests. These results highlight the potential of foundation models as scalable and adaptive tools for financial forecasting-capable of delivering strong performance in dynamic market environments when paired with targeted fine-tuning strategies.

Complexity vs Empirical Score

  • Math Complexity: 5.0/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper applies advanced deep learning techniques (transformer architecture, incremental learning) but primarily focuses on empirical evaluation, statistical tests, and extensive backtesting across 21 global indices.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Evaluate TimesFM for Volatility Forecasting"] --> B["Data: S&P 500 & NASDAQ<br>Realized Volatility Series"]
  B --> C["Methodology: Three-Stage Approach"]
  
  C --> D["1. Zero-Shot<br>Pretrained TimesFM"]
  C --> E["2. Fine-Tuned<br>Incremental Learning"]
  C --> F["3. Baseline<br>Standard Econometrics"]
  
  D --> G["Testing & Comparison"]
  E --> G
  F --> G
  
  G --> H{"Evaluation"}
  H --> I["Forecast Accuracy<br>vs Benchmarks"]
  H --> J["Statistical Tests<br>DM & GW"]
  
  I --> K["Key Findings"]
  J --> K
  
  K --> L["• Incremental Fine-Tuning Essential"]
  K --> M["• Outperforms Econometrics"]
  K --> N["• Scalable for Financial Risk"]
  
  style A fill:#e1f5fe
  style B fill:#fff3e0
  style C fill:#f3e5f5
  style K fill:#e8f5e8

Cited References (71)

  1. A machine learning approach to volatility forecasting (2023)
  2. Neural network–based financial volatility forecasting: A systematic review (2022)
  3. Volatility forecast using hybrid neural network models (2014)
  4. GARCH based artificial neural networks in forecasting conditional variance of stock returns (2014)
  5. Combining conditional volatility forecasts using neural networks: an application to the EMS exchange rates (1999)
  6. Artificial neural networks for realized volatility prediction in cryptocurrency time series (2019)
  7. Evolving hybrid neural fuzzy network for realized volatility forecasting with jumps (2014)
  8. Forecasting realized volatility: HAR against Principal Components Combining, neural networks and GARCH (2017)
and 63 more
  1. Evolving fuzzy-GARCH approach for financial volatility modeling and forecasting (2016)
  2. A hybrid modeling approach for forecasting the volatility of S&P 500 index return (2012)
  3. Using neural networks for forecasting volatility of S&P 500 Index futures prices (2004)
  4. Forecast combining with neural networks (1996)
  5. HAR volatility modelling with heterogeneous leverage and jumps (2009)
  6. Volatility forecasting with machine learning and intraday commonality (2024)
  7. Stock market volatility forecasting: Do we need high-frequency data? (2021)
  8. Volatility forecasting using deep recurrent neural networks as GARCH models (2023)
  9. Comparing predictive accuracy (2002)
  10. A forecast comparison of volatility models: does anything beat a GARCH (1, 1)? (2005)
  11. Power and bipower variation with stochastic volatility and jumps (2004)
  12. Roughing It Up: Including Jump Components in the Measurement, Modeling, and Forecasting of Return Volatility (2007)
  13. Good volatility, bad volatility: Signed jumps and the persistence of volatility (2015)
  14. Exploiting the errors: A simple approach for improved volatility forecasting (2016)
  15. A simple approximate long-memory model of realized volatility (2009)
  16. Realized GARCH: a joint model for returns and realized measures of volatility (2012)
  17. Volatility forecast comparison using imperfect volatility proxies (2011)
  18. Discrete-time volatility forecasting with persistent leverage effect and the link with continuous-time volatility modeling (2012)
  19. Forecasting stock market volatility using realized GARCH model: International evidence (2016)
  20. The model confidence set (2011)
  21. Tests of conditional predictive ability (2006)
  22. A decoder-only foundation model for time-series forecasting (2024)
  23. Using LSTM and GRU neural network methods for traffic flow prediction (2016)
  24. Time-Series Foundation AI Model for Value-at-Risk Forecasting — → 2410.11773
  25. Estimating quadratic variation using realized variance (2002)
  26. A time series is worth 64 words: Long-term forecasting with transformers (2022) — link
  27. Chronos: Learning the language of time series (2024)
  28. Lag-llama: Towards foundation models for time series forecasting (2023)
  29. Visionts: Visual masked autoencoders are free-lunch zero-shot time series forecasters (2024) — link
  30. Foundation models for time series analysis: A tutorial and survey (2024)
  31. Unified training of universal time series forecasting transformers (2024)
  32. Tiny time mixers (ttms): Fast pre-trained models for enhanced zero/few-shot forecasting of multivariate time series (2024)
  33. The M4 Competition: 100,000 time series and 61 forecasting methods (2020)
  34. A review of generalized zero-shot learning methods (2022)
  35. DeepAR: Probabilistic forecasting with autoregressive recurrent networks (2020)
  36. Monash time series forecasting archive (2021) — link
  37. N-BEATS: Neural basis expansion analysis for interpretable time series forecasting (2019) — link
  38. Informer: Beyond efficient transformer for long sequence time-series forecasting (2021)
  39. Gpt (generative pre-trained transformer)–a comprehensive review on enabling technologies, potential applications, emerging challenges, and future directions (2024)
  40. Mamba4Cast: Efficient Zero-Shot Time Series Forecasting with State Space Models (2024) — link
  41. Timedit: General-purpose diffusion transformers for time series foundation model (2024) — link
  42. Masked autoencoders are scalable vision learners (2022)
  43. Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation (1982)
  44. Generalized autoregressive conditional heteroskedasticity (1986)
  45. Modeling and predicting the CBOE market volatility index (2014)
  46. Novel volatility forecasting using deep learning–long short term memory recurrent neural networks (2019)
  47. The benefits of bagging for forecast models of realized volatility (2010)
  48. Introducing NBEATSx to realized volatility forecasting (2024)
  49. Machine learning for realised volatility forecasting (2020)
  50. An artificial neural network-GARCH model for international stock return volatility (1997)
  51. Realized volatility forecasting with neural networks (2020)
  52. Trading volume and realized volatility forecasting: Evidence from the China stock market (2023)
  53. Forecasting realised volatility using ARFIMA and HAR models (2019)
  54. Financial returns modelled by the product of two stochastic processes-a study of the daily sugar prices 1961-75 (1982)
  55. Lassoing the HAR model: A model selection perspective on realized volatility dynamics (2016)
  56. The impact of sentiment and attention measures on stock market volatility (2020)
  57. Building news measures from textual data and an application to volatility forecasting (2017)
  58. Forecasting of realised volatility with the random forests algorithm (2018)
  59. Stock market volatility: Identifying major drivers and the nature of their impact (2015)
  60. Data-driven neural architecture learning for financial time-series forecasting (2019) — link
  61. Deep adaptive input normalization for time series forecasting (2019)
  62. Deeplob: Deep convolutional neural networks for limit order books (2019)
  63. Using machine learning to predict realized variance (2020)

1 of 71 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.